Most sources will take a particular stance or are produced from the view point of a particular interest. A review of articles, analyses, etc will likely not represent an objective standard of analysis. Other analysts are responding to the critical intelligence requirements (CIR) of a decision maker, or customer. Articles often serve the purposes of a particular interest, individual experience, or cater to sensational (recent and vivid) pieces of information. A review of such sources may produce some understanding of the issues and entities that individuals are paying attention to, yet it may fail to present a objective foundation of analysis.
Cause and Effect vs. Correlation:
Correlation is one of the most powerful tools of indication and prediction; however, correlation presents one of the significant confound dangers to analysis. Confounds are variables that vary systematically with the measured (target) aspects of analysis and often provide alternative explanation for cause and effect relationships. This is to say, two components of analysis may vary in a correlation to each other without having any causal relationship to each other. One potential explanation for this is the "third variable confound." To be sure, it is sometimes the case that two characteristics of analysis may react in a similar manner to an influence but do not react in accordance of the intrinsic nature of the other. Instead, both of the characteristics share a causal relationship to a third characteristic. The confound exists when one of the original characteristics exists in the absence of the "third characteristic." We then expect to see the second "correlated" characteristic, but it is not present in the absence of the third "causal" characteristic.
Assumptive Biases and Artifacts:
An analyst inevitably brings assumptions and heuristics to their analysis. Even a "common sense bias" may taint the analysis. This is to say, that some things are taken for granted because they make sense. We may conclude that the influence or efficacy of a separatist movement is based on its popular support. This is not always the case. A separatist movement may serve an important purpose for or represent political ideology more favorable to other more power actors in the region who fund and support the movement. Membership does not always equal popular support. Many separate movement in Africa raid villages and kidnap children to man their growing armies of separatists. Over structuring analysis may produce results consistent with the original hypothesis, yet fail to truly capture a broad generalizable analysis of the intelligence target. Such results are artifacts. If the contextual structure of an analysis is too specific it runs the danger of creating such artifacts. To be sure, narrowing specific criteria to analyze a broad question such as "the role of NSA in Sub-Saharan Africa" will produce analysis that is effective in only very specific circumstances, and will not serve broad analytical purposes.
The Atypical Correlate
In many historical studies it has been the case that the most reliable correlates have demonstrated little apparent connection to the hypothesis being studied. Whether by accident, imagination, or diligent study, researchers have uncovered lateral correlates that often represent the most indicative and reliable information in their analysis. Sometimes this correlate has been found by identifying "third variable confounds." These "diamond in the rough" patterns are sometimes difficult to find, but open-minded diligent analysis may uncover such intelligence gems.
The Algorithm and the Danger of Too Much Information:
Malcolm Gladwell, in his book Blink, describes how people sometimes use too much information to make a decision for a particular circumstance. Gladwell describes how the Cook County Hospital attempted to deal with the problem of predicting heart attacks in patients complaining of chest pain. A 4 factor algorithm, developed by Lee Goldman, proved significantly more effective than a doctor’s (expert’s) diagnosis (opinion/assessment). Gladwell’s conclusion is that we sometimes use too much information to produce our assessments of particular situations. We all know that family histories of heart attacks greatly increase one’s risk of having a heart attack in one’s life, but does it predict whether someone will have a heart attack over the next 72 hours? The answer is no, a family history simply presents an increased risk, but does not help identify the cause of chest pain, at least not significantly. Chest pain can be the result of several different causations (confounds), and an individual is likely to have many bouts of chest pain over their life, especially in later life. Genetic predisposition increases the chance of heart attacks primarily in later life. Therefore, genetic predisposition is a poor indicator for any 72 hour period. However, electrocardiographic (ECG) evidence, presence of fluid in the lungs, unstable angina, and systolic blood pressure in excess of 100mm Hg are good indicators of a looming heart attack threatening to strike in the next 72 hours. This means that any nurse or even you or I can be quickly trained to speedily predict a looming heart attack and outperform the average expert (cardiologist). The director of the Cook County Hospital, Dr. Brandon Reilly, found that the experts (doctors) were considering indicators such as family history, diabetes, cholesterol, etc which are long-term risk factors for heart attacks, but shed little light on the next 72 hours, and produce assessments with lower accuracy than those exclusively considering the 4 factor algorithm. The individual using the 4 factor algorithm is “high speed” going directly for the most effective indicators, and “low drag” ignoring information that is cumbersome and complicates the immediate need.
The Structured Analysis Approach:
Distilling the vast amount of information down into the most indicative and reliable pieces (indicators) is of paramount concern when tackling a query as broad as "determining the role of non-state actors in Sub-Saharan Africa." Once key indicative aspects of a non-state actor are identified, a general algorithmic model can be created for analyzing non-state actors. Anomaly-centric analysis can then be conducted to identify potential flaws in the algorithmic model. Such analysis will identify non-state actors that don't fit the model and can help refine the model. Using such a structured approach can combat analyst biases and prevent the danger of too much information.
Executive Summary
The Dilemma:
Most sources will take a particular stance or are produced from the view point of a particular interest. A review of articles, analyses, etc will likely not represent an objective standard of analysis. Other analysts are responding to the critical intelligence requirements (CIR) of a decision maker, or customer. Articles often serve the purposes of a particular interest, individual experience, or cater to sensational (recent and vivid) pieces of information. A review of such sources may produce some understanding of the issues and entities that individuals are paying attention to, yet it may fail to present a objective foundation of analysis.Cause and Effect vs. Correlation:
Correlation is one of the most powerful tools of indication and prediction; however, correlation presents one of the significant confound dangers to analysis. Confounds are variables that vary systematically with the measured (target) aspects of analysis and often provide alternative explanation for cause and effect relationships. This is to say, two components of analysis may vary in a correlation to each other without having any causal relationship to each other. One potential explanation for this is the "third variable confound." To be sure, it is sometimes the case that two characteristics of analysis may react in a similar manner to an influence but do not react in accordance of the intrinsic nature of the other. Instead, both of the characteristics share a causal relationship to a third characteristic. The confound exists when one of the original characteristics exists in the absence of the "third characteristic." We then expect to see the second "correlated" characteristic, but it is not present in the absence of the third "causal" characteristic.Assumptive Biases and Artifacts:
An analyst inevitably brings assumptions and heuristics to their analysis. Even a "common sense bias" may taint the analysis. This is to say, that some things are taken for granted because they make sense. We may conclude that the influence or efficacy of a separatist movement is based on its popular support. This is not always the case. A separatist movement may serve an important purpose for or represent political ideology more favorable to other more power actors in the region who fund and support the movement. Membership does not always equal popular support. Many separate movement in Africa raid villages and kidnap children to man their growing armies of separatists. Over structuring analysis may produce results consistent with the original hypothesis, yet fail to truly capture a broad generalizable analysis of the intelligence target. Such results are artifacts. If the contextual structure of an analysis is too specific it runs the danger of creating such artifacts. To be sure, narrowing specific criteria to analyze a broad question such as "the role of NSA in Sub-Saharan Africa" will produce analysis that is effective in only very specific circumstances, and will not serve broad analytical purposes.The Atypical Correlate
In many historical studies it has been the case that the most reliable correlates have demonstrated little apparent connection to the hypothesis being studied. Whether by accident, imagination, or diligent study, researchers have uncovered lateral correlates that often represent the most indicative and reliable information in their analysis. Sometimes this correlate has been found by identifying "third variable confounds." These "diamond in the rough" patterns are sometimes difficult to find, but open-minded diligent analysis may uncover such intelligence gems.The Algorithm and the Danger of Too Much Information:
Malcolm Gladwell, in his book Blink, describes how people sometimes use too much information to make a decision for a particular circumstance. Gladwell describes how the Cook County Hospital attempted to deal with the problem of predicting heart attacks in patients complaining of chest pain. A 4 factor algorithm, developed by Lee Goldman, proved significantly more effective than a doctor’s (expert’s) diagnosis (opinion/assessment). Gladwell’s conclusion is that we sometimes use too much information to produce our assessments of particular situations. We all know that family histories of heart attacks greatly increase one’s risk of having a heart attack in one’s life, but does it predict whether someone will have a heart attack over the next 72 hours? The answer is no, a family history simply presents an increased risk, but does not help identify the cause of chest pain, at least not significantly. Chest pain can be the result of several different causations (confounds), and an individual is likely to have many bouts of chest pain over their life, especially in later life. Genetic predisposition increases the chance of heart attacks primarily in later life. Therefore, genetic predisposition is a poor indicator for any 72 hour period. However, electrocardiographic (ECG) evidence, presence of fluid in the lungs, unstable angina, and systolic blood pressure in excess of 100mm Hg are good indicators of a looming heart attack threatening to strike in the next 72 hours. This means that any nurse or even you or I can be quickly trained to speedily predict a looming heart attack and outperform the average expert (cardiologist). The director of the Cook County Hospital, Dr. Brandon Reilly, found that the experts (doctors) were considering indicators such as family history, diabetes, cholesterol, etc which are long-term risk factors for heart attacks, but shed little light on the next 72 hours, and produce assessments with lower accuracy than those exclusively considering the 4 factor algorithm. The individual using the 4 factor algorithm is “high speed” going directly for the most effective indicators, and “low drag” ignoring information that is cumbersome and complicates the immediate need.The Structured Analysis Approach:
Distilling the vast amount of information down into the most indicative and reliable pieces (indicators) is of paramount concern when tackling a query as broad as "determining the role of non-state actors in Sub-Saharan Africa." Once key indicative aspects of a non-state actor are identified, a general algorithmic model can be created for analyzing non-state actors. Anomaly-centric analysis can then be conducted to identify potential flaws in the algorithmic model. Such analysis will identify non-state actors that don't fit the model and can help refine the model. Using such a structured approach can combat analyst biases and prevent the danger of too much information.